directaccess: use cached filteredrevs Before this patch we were calling directly repoview.computehidden(repo) to compute the revisions visible with direct access, without going through the caching mechanism for the filtered revisions. There was two issues with that: (1) Performance: We were not leverating the cached values of the 'visible' revs (2) Stability: If there were to be a cache inconsistency with the computation of 'visible' we would crash in the branchmap consistency check partial.validfor. Consider the scenario of rebase with bookmarks: - when we delete a bookmark on an obsolete changeset (like what rebase does when moving the bookmark after rebasing the changesets) - then this changes the value returned by repoview.computehidden(repo) as bookmarks are used as dynamic blockers in repoview.computehidden(repo) - as of now, we don't invalidate the cache in the case of bookmark change - if we have a cached value from before the bookmark change, repoview.filterrevs(repo, 'visible') considers the cached value correct and returns something different than repoview.computehidden(repo) - in turn, if we use repoview.computehidden(repo) in directaccess, the subset relationship is broken and the cache consistency assertion (parial.validfor) fails if branchmap.updatecache is called in this time window This patch leverages the caching infrastructure in place to speed up the computation of the filteredrevs for visible-directaccess-nowarn and visible-directaccess-warn. Incidentally it prevents the bug discussed in (2) from crashing when running a rebase with a bookmark. Note that there still needs to be a fix in core for the case discussed in (2). The test for this side of the fix (not core's fix for (2) is very hard to implement without introducing a lot of dependencies and does not belong here. It is much easier to have the test of the fix for the scenario (2) in core along with the fix.

inhibit: improve performance of transaction wrapping Before this patch, transaction wrapping was the most expensive part of inhibit computation wise. This patch changes the revset that we use in the transaction wrapping to make it ~50x faster to compute: revset #0: obsolete() - hidden() 0) wall 0.000214 comb 0.000000 user 0.000000 sys 0.000000 (best of 11209) vs revset #0: (not hidden()) and obsolete() 0) wall 0.010965 comb 0.010000 user 0.010000 sys 0.000000 (best of 237)

evolve: extract the code computing dependencies in a separate function The code to compute dependencies between unstable changeset can be reused to compute the next reveset. This patch extracts it in its own function to make it reusable.

directaccess: add mechanism to load directaccess after some other extensions directaccess needs to load after some extensions to avoid interfering with them. This patch adds a mechanism to specify what extension directaccess needs to load after.

evolve: add another test for evolve --rev This patch adds one test for three conditions not covered before: - evolve --rev on a stack with obsolete commits without successors - evolve --rev on two sets of stacks in parallel - evolve --rev on a set of stack partially solvable (because all of the commits of the stack are not specified with --rev)

evolve: add a more complex test for evolve --rev The previous tests of evolve --rev would have passed with an ordering of revs by increasing revision numbers. This patch adds a test that would fail if that was what we were doing and therefore tests the implementation better.

evolve: add ordering of the revisions for evolve --rev When running evolve --rev we want to process the revisions in an optimal fashion to solve the maximum amount of trouble in the minimum number of steps. This patch adds a step to evolve --rev to order the revision before solving the troubles. A simple test is added to cover a basic case.

test: adapt to change in mercurial core Bundle2 is one by default and more capability have been added.

evolve: add directaccess to the setup.py directaccess was missing from the list so that when we were building evolve it was not contained in the build.

evolve: don't crash on singled out revisions Before this patch, we were crashing on unsolvable revisions when using evolve --rev. This patch removes the crash and prints a warning when we encounter such revision.

test: adapt test to less frequent branch warning The 701df761aa94 changeset in mercurial core made the branch warning issue only for the first branch created. We adapt the tests to reflect this change.

tests: adapt summary output to 6084926366b9 The 'summary' commands recently gained phases related output, 6084926366b9 put it into a "final" state. We adapt the test to this new style.

evolve: small refactoring of the uncommit function We introduce a new variable to simplify the implementation of hg uncommit --interactive.

evolve: improve error message We add a new line and change the wording from 'revsets' to 'revisions'.

evolve: move return statement at the right level When no trouble was found with the --rev option, the code of the evolve function would continue running and would solve unwanted troubles. This patch makes sure it does not happen.

evolve: remove extra whitespace This patch removes a whitespace at the end of a line.

directaccess: don't crash when evolve is not loaded Before this patch, when evolve was not loaded direct access was crashing. This patch catches the error and prevents the crash.

inhbit: don't crash on commit with no changes Before this patch inhibit would crash when running hg amend with no changes. This patch fixes this case and adds a test to prevent regression.

evolve: add a test for prune --keep -r . with active bookmark We recently changed the behavior of prune to make sure that the active bookmark stays active after prune -r . This patch adds a test for the case of prune --keep -r . that does not take the same code path.

Merge with stable

inhibit: add test to ensure that --hidden is working with inhibit There was a concern with the transaction hook in inhibit and its potentially devastating effect when used with --hidden. The transaction hook in inhibit hides all the visible & obsolete commits when the transaction ends using the visible() and the obsolete() revset. --hidden does not change the visible() revset so it is fine. If this changes in the future and --hidden actually ends up impacting the visible() revset, this test will catch it and prevent inhibit from putting inhibition markers on all of the obsolete changesets.

prune: with active bookmark should stay active hg prune with an active bookmark was deactivating the bookmark after the prune. This patch makes sure that if a bookmark was active before a "prune" operation (except prune -B) then the bookmark stays activate afterwards.

merge with stable

evolve: migrate off of now-dead util.any

inhibit: create direct access extension Since we want to use direct access without necessarily using inhibit, this patch separates both extensions. Inhibit depends on direct access and we add a test to check that it complains if that is not the case.

inhibit: handle inhibit marker on stripped revision If a revision disappear from the repo, we should not crash.

update README

evolve: don't use python sets on top of revset for evolve --rev In --rev we were wrapping the revsets in python sets. This is wrong as the use python sets force the wrapper revision set to be computed earlier than it has to and have no order guarantee. Removing this wrapping changes a test because the ordering does not appear to be the same but the end result is the same.

evolve: current bookmark wrongly moving during prune (issue4559) Before this patch: prune -r <otherbookmark> also moved the current bookmark eventhough it was not related to the pruned changes. This patch fixes it and adds a test to catch regressions.

inhibit: direct access with and without warning on a per command basis We introduce a new filtername visibile-directaccess-nowarn to enable direct access with no warning on a per command basis. The motivation behing this change is to display warning when attempting direct access in destructive commands.